Expand and qualify the cellsIdentify and cryopreserve aliquotsRecord locations and retrieve as needed

1. What a cell bank is (the short version)

The glossary of ICH Q5D describes a cell bank as a collection of suitable containers with uniform contents, stored under defined conditions, in which each container is an aliquot of one pool of cells Sourced [Source 1].

• Why keep a stock: as an advantage of using passaged cells, Q5D lists that every production lot can start from a common, characterised starting materialSourced [Source 1, section 2.2].
• How far it is needed: Part IV of EU GMP, for ATMPs, "recommends" a bank system for allogeneic products where donor and patient need not match, but says that establishing one is "not mandatory"Sourced [Source 2, 8.10]. For autologous products (made from the patient's own cells), the starting material is that patient's cells, so the idea of a bank applies less readily (see our explainer on autologous and allogeneic).
• What it means for a materials engineer: once made, a bank can be drawn on at every manufacture, but it is also a raw material that cannot be remade. Containers, cryopreservation solution, tubing, cooling, storage units, closed systems: a mistake in choosing any of them reaches every later production lot (commentary).

2. Two-tier banks — master and working, and how they look for cell products

Q5D calls the two-tier system, in which a working cell bank (WCB) is made from a master cell bank (MCB), "generally accepted as the most practical approach"Sourced [Source 1, section 2.2.1].

The reasons for two tiers: use-up and remaking

Q5D says the MCB is made first and the WCB is made from the MCB "as needed". The WCB is used directly in production, and the MCB is drawn down a little each timeSourced [Source 1, section 2.2.1]. It adds that if only a few containers are needed each year, a one-tier system with an MCB alone could "in principle" be usedSourced [Source 1, section 2.2.1]. For evaluating a bank system, EU Part IV asks for validation beyond the passage number or population doublings reached in routine manufactureSourced [Source 2, glossary].

For cell products, the picture differs a little from document to document

All Sourced (Sources 1, 2, 4 and 5). The right-hand column is commentary from this article.

A real iPS cell stock: what failed in testing

In presenting its ten-year summary paper on iPS cell stocks for regenerative medicine (Med, dated 16 November 2022), the CiRA Foundation published the followingSourced [Source 6].

• iPS cell lines were established from peripheral blood or cord blood of 7 HLA-homozygous donors, and after quality testing 27 lines were selected and are offered
• Manufacture was changed in two steps: single-cell cloning was introduced when making the SCS from the PCS (Method 2), and colony picking at the PCS stage was reduced to shorten manufacturing time and passage number (Method 3)
• The main reasons lines were not selected were residual episomal vector, mutations in cancer-related genes and copy-number or structural abnormalities found by genome analysis, undifferentiated-marker results (low TRA-1-60 expression), and karyotype abnormalities. In the PCS of one donor, nearly 40% showed karyotype abnormalities
• No line was excluded by specification tests such as contamination testing, sterility testing included

What can be read from this (commentary): the main cause of banks being rejected was not contamination but the nature of the cells themselves. Bank testing brings to mind sterility, mycoplasma and viruses, but for iPS cells genetic quality (genome, karyotype, residual vector) is at the centre of selection. That no line fell to sterility testing can also be read as contamination prevention working on the making side (consistent with the Q5D principle). This is, however, the result of one project and cannot be generalised to other cells or facilities.

3. How a bank is made, and what must be protected

Section 2.2.2 of Q5D sets out concretely how a bank is madeSourced [Source 1].

• Divide from one pool: culture is expanded step by step, the cells from several culture vessels are combined into one pool, then dispensed into sterile containers and sealed. This makes the contents of all containers uniform
• What to record: the size of the bank, the containers (vials, ampoules and so on) and closures, the method of preparation (including the cryoprotectant and medium), and the conditions of freezing and storage. Labels must keep their information through storage, keeping and recovery
• Freezing and storage: animal cells are frozen in medium containing a cryoprotectant, in the sealed container, under defined controlled conditions, and moved to the vapor or liquid phase of liquid nitrogen or an equivalent ultra-low temperature
• Preparing for disasters: so that fire, power failure or human error do not wipe everything out, it gives examples such as spreading across several freezers, emergency power, automatic refilling of liquid nitrogen, keeping part of the bank at a remote site, and the ability to remake the MCB
• Where the cells' "age" starts: the in vitro age of cells in manufacture is counted from the time a container of the MCB is thawed

The EU's Part IV is more concrete about storage and useSourced [Source 2, 8.15 to 8.21].

All Sourced (EU GMP Part IV [Source 2]).

The Japanese notification likewise asks that, for storage and transport, appropriate stability tests based on cell viability and potency be run with full regard to the storage and distribution period and the storage form, and the storage method and expiry set accordingly; that where freezing, storage and thawing are used, it be confirmed that freezing and thawing have no effect; and that transport containers and procedures (including temperature control) be defined and justified Sourced [Source 5, chapter 2, part 1, item 4]. It also says that where the cryopreservation period during manufacture or the cell culture period is long, sterility is to be confirmed, for example by sterility testing at regular intervalsSourced [Source 5, chapter 2, part 2, item 6].

4. Bank testing — what is checked, and how much

Q5D states the aim of testing as confirming the identity, purity and suitability of the cell substrate for manufacture. The frequency is organised as identity and purity once per MCB, and stability once per product registered, and for a WCB purity and limited identity once eachSourced [Source 1, section 2.3].

All Sourced (Q5D [Source 1], FDA guidance [Source 4], MHLW notification [Source 5]). The arrangement of the table is this article's and does not cover every item in each document.

From the discussion of the International Stem Cell Banking Initiative (ISCBI)

The report of an ISCBI workshop held in 2017 raised the following points on quality control of pluripotent stem cell banksSourced [Source 7].

• It is known that conventional sterility tests cannot detect every microorganism that can contaminate a cell culture, and new molecular methods for detecting bacteria and fungi are emerging
• The blood-borne viruses tested for in transplantation may not be enough for cell therapy products, and a wider range of viruses needs to be considered based on risk assessment of the donor and raw materials
• The teratoma assay in mice, used to confirm pluripotency, takes 6 to 12 weeks, uses animals, lacks reproducibility and is expensive, so it is not desirable as a quality control test
• There was discussion that, if genome sequencing is done, it should be done on both the cell line and the final product
• On cost, a presenter estimated that testing a research-grade cell bank costs US$80,000 to US$200,000 (depending on the tests), with about US$150,000 more for preparing the bank. Clinical grade needs validated test methods and costs still more

All Sourced (ISCBI workshop report [Source 7]). The costs are a presenter's rough estimate, figures for research-grade cell lines in 2017.

5. Our calculation: vials used for testing, DMSO concentration and wetted area

6. A materials engineer's view (1): cryopreservation solutions and cooling — container shape sets the cooling rate

What happens to cells when they freeze

Hunt's review (2011) describes the standard procedure for freezing cells as cooling slowly in the presence of a cryoprotectant to avoid damage from ice inside the cell. Dimethyl sulfoxide (DMSO) is widely used as the cryoprotectant, and controlled-rate freezing at 1 to 2 °C per minute with rapid thawing is the standardSourced [Source 8].

• Cooled slowly: ice forms first outside the cells, and as water turns to ice the solute concentration of the remaining liquid rises. Both ice and high solute concentration harm cells. A protectant eases the damage by reducing the amount of ice formed at a given temperature and limiting the rise in solute concentrationSourced
• Cooled fast: the risk of ice forming inside the cells risesSourced. In clumps of cells (colonies), experiments have shown that intracellular ice may propagate to neighbouring cells, which is discussed (as a hypothesis) as one reason human pluripotent stem cells recover poorly
• Vitrification: a method of making the liquid set into a glass, either with a protectant concentrated enough to prevent ice or with extremely fast cooling. High protectant concentrations raise toxicity problems, and the method of plunging very fine tubes directly into liquid nitrogen is not suited to mass production or GMP, the review saysSourced

Show the detailed vector diagram and its source notes

What is in a cryopreservation solution — from "DMSO plus serum" to defined solutions

The cryoprotective action of DMSO is an old technology, reported in Nature in 1959Sourced [Source 15]. Hunt's review, however, says DMSO toxicity depends on time, temperature and concentration and differs with cell type. For infusion of haematopoietic stem cells, adverse reactions have been reported when DMSO was not removed after thawing, and efforts have been made to wash (using an osmotic buffer containing dextran) and to lower the concentration to 5%, even 2%Sourced [Source 8]. Alternative protectants and additives under study include hydroxyethyl starch, trehalose, polyvinylpyrrolidone (PVP), methylcellulose, polyethylene glycol (PEG), polyampholytes (such as poly-L-lysine) and sericin, but at the time of that review most were at the laboratory stageSourced [Source 8].

All Sourced (Hunt 2011 [Source 8], Dobruskin 2025 [Source 11], EU GMP Part IV [Source 2]). Hunt's review dates from 2011, and commercial products and the state of research may have changed.

Pluripotent stem cells were cells that were hard to freeze

Hunt's review introduces early work on human ES cells in which standard slow freezing gave low recovery (colony recovery of roughly 5% to 16% in one case) while vitrification gave over 75%Sourced [Source 8]. But because vitrification plunges a tiny volume of liquid directly into liquid nitrogen, it was said to have problems for mass production and for preventing contamination. The same review also lists later examples in which slow freezing with adjusted conditions raised recovery.

• Seeding: cooling at under 1.8 °C per minute in 10% DMSO with seeding gave about 80% viability (low without seeding)Sourced
• Programmed cooling: 1 °C per minute, seeding at -7 °C and one-step addition of 10% DMSO gave over 50% reattachment and recoverySourced
• ROCK inhibitor: adding Y-27632 to the medium for 24 hours after thawing raised human ES cell viability about 10-fold. A similar effect has been reported for iPS cellsSourced

All Sourced (results of other studies introduced in Hunt 2011 [Source 8]). The cell lines and measurement methods differ, so the numbers cannot be compared with one another.

7. A materials engineer's view (2): containers and cold storage — "do not crack, do not leak, do not contaminate"

Where to store — liquid or vapor phase of liquid nitrogen, or mechanical

Hunt's review explains that long-term bank storage requires keeping below the glass transition temperature to stop molecular motion, and that this has been done in the liquid or vapor phase of liquid nitrogen. It also introduces results that mechanical freezers holding steadily below -135 °C have become reliable and that, in studies on stem cells, no significant difference from liquid nitrogen was found for up to about 5 yearsSourced [Source 8].

What prompted a rethink of liquid-phase storage was hepatitis B in 1990s Britain. In the report by Tedder et al., six patients with blood cancers and the like developed acute hepatitis B over 25 months, and the bone marrow or peripheral blood stem cells of all of them had been stored in the same freezing tank. They concluded that a leak in the freezing bag holding the marrow of the first infected patient contaminated the tank and its contents, and that infection spread through transplantationSourced [Source 10]. The report stressed that the primary container of frozen tissue must be sealed so that nothing is exchanged between specimen and liquid nitrogenSourced [Source 10]. Hunt's review says clinical material then moved to vapor-phase storage in liquid nitrogen [Source 8]. EU Part IV likewise gives, as an example, banks stored as sealed containers in the vapor phase of liquid nitrogen (8.15) [Source 2].

Materials of containers and connections — what the primary sources confirm

Everything in the "what the primary sources say" column is Sourced (Hunt 2011 [Source 8], Woods et al. 2011 [Source 9]). The paper by Woods et al. is an early validation of one commercial product (a closed cryovial), and we did not confirm whether there is a conflict of interest. The right-hand column is commentary from this article.

Woods et al. compared this closed vial with a standard screw-cap vial in a drop test from a height of 1 m in the frozen state, a dye ingress test, and a microbial ingress test by an outside contract laboratory (immersion in a suspension of the very small bacterium Brevundimonas diminuta under a pressure difference). They report that after freezing to -196 °C, storage, shipping and thawing, no dye ingress was detected and sterility was keptSourced [Source 9]. This is one example of confirming container performance before and after lowering the temperature. It is, however, a result for one container and does not show the performance of containers in general.

8. Closed systems and automation — what counts as "closed"

Most cell products cannot be sterilised in their final form. EU Part IV says that since the great majority of ATMPs cannot be terminally sterilised, manufacture is carried out aseptically (under conditions that prevent contamination)Sourced [Source 2, 9.39]. That is why a closed system, which does not expose the product to room air, is a central idea in cell manufacture. We set out its definition and the background environment it requires from the primary sources.

The definition of "closed system" differs a little between documents

All Sourced (EU GMP Part IV [Source 2], Annex 1 [Source 3]). In a concept paper dated March 2025, the EMA set out a plan to revise Part IV to align with the current Annex 1, saying that the revision would clarify expectations for qualification and control of cleanrooms and closed systems (isolators and RABS) and accommodate new technologies absent from the current text, such as automation and (closed) single-use systemsSourced [Source 16]. We could not confirm that a revised version has been finalised as of this survey (October 2026)Not yet confirmed.

The Annex 1 view of closed systems

• A closed system is designed to reduce the risk of contamination by microorganisms, particles and chemicals from the surrounding environment, and to "reduce the need for manual intervention" (8.127)
• Sterility of every surface that touches the product is kept. Where equipment tubing and the like are connected to the flow path after sterilising final filtration, the design allows aseptic connection, for example with intrinsically sterile connection devices (8.128)
• The method of ensuring the integrity of components used in aseptic connection is defined, and where there is a risk of losing sterility, system integrity testing is considered. Supplier assessment includes collecting data on failure modes that lose sterility (8.129)
• If integrity could be compromised in aseptic manufacture, the system is placed in Grade A. It may be used in a lower grade if integrity is shown to be kept at each use (by pressure testing or monitoring) (8.130)

All Sourced (EU GMP Annex 1 [Source 3]). Annex 1 is a document for sterile medicinal products in general. Other documents in EudraLex Volume 4 do not apply to ATMPs unless explicitly referenced (see our explainer on GMP).

Automated equipment — Part IV chapter 17

EU Part IV has a chapter on "automated manufacture of ATMPs". Only the key points are given here (for the treatment of the CE mark and the like, see our explainer on CDMOs)Sourced [Source 2].

• Using automated equipment can make some GMP requirements easier to meet and may bring advantages for product quality. But responsibility for quality lies with the ATMP manufacturer (17.11 to 17.12)
• Equipment is used only under conditions that keep aseptic processing. For example: validation of cleaning, sterilisation of reused parts, and confirmation of equipment integrity by pressure-hold tests, leak tests and the like (17.21)
• A room where closed systems are used is at least Grade D, and moving materials in and out of the equipment is a critical step. Validated procedures are needed (17.24)
• Because in-process controls are limited in continuous closed processing, continuous monitoring of critical process parameters is done where technically possible (17.27)
• Media fills (simulation of aseptic processing with culture medium) are run. The text recommends a frequency of "bi-annual" (wording that can be read as twice a year or every two years), adjustable according to risk (17.28)

As a real example, Matsumoto et al. cultured iPS cell-derived retinal pigment epithelium (RPE) cell sheets automatically in equipment with closed culture modules. They report that the machine-cultured sheets were of equivalent quality to hand-made sheets in comparisons of morphology and biological propertiesSourced [Source 12]. The paper lists as problems of manual work variation in quality with operator skill, the labour required and cost. This is, of course, a comparison for one process and does not show the effect of automation in general.

Sealed source bagTransfer through aseptically connected tubingSealed receiving bag

9. A materials engineer's view (3): the weak point of a closed system is the joint

10. Validation — what backs trust in the bank and the process

Trust in banks and closed systems is backed not by a single test but by an accumulation of checks. We sort out, from the primary sources, what is checked.

All Sourced (Q5D [Source 1], Part IV [Source 2], Annex 1 [Source 3], MHLW notification [Source 5], Hunt [Source 8], Woods et al. [Source 9]). The arrangement of the table is this article's.

Beyond the bank comes the expansion culture, differentiation and formulation that begin with thawing the WCB. Q5D counts the in vitro age of cells in manufacture from the time a container of the MCB is thawedSourced [Source 1]. For culture vessels, stirring and oxygen transfer, see our explainer on bioreactors; for the clauses of GMP, the GMP explainer; for change control, the QbD and CMC explainer; and for the structure of contract manufacturing, the CDMO explainer.

11. Open issues, and what is still undecided

(1) We could not confirm clinical use of DMSO-free freezing

The 2025 review says research on DMSO-free cryopreservation solutions is advancing, but there are not yet clinical trialsSourced [Source 11]. Changing the freezing solution of a bank is a change to the process, so an impact assessment of the change (see the QbD and CMC explainer) would be needed (commentary).

(2) We could not confirm a unified standard for evaluating container integrity at low temperature

Annex 1 asks for attention to the integrity of components exposed to freezing and thawing [Source 3], but within the scope of this survey we could not confirm a unified standard that also sets the test methods. The tests by Woods et al. are an early validation of one container, and we have not confirmed whether there is a conflict of interestNot yet confirmed.

(3) Resin property values (such as the transition temperature of EVA) rest on a statement in a single paper

The EVA transition temperature in Fig. 4 (beginning near -15 °C) is a statement by Woods et al. [Source 9]. It may change with the vinyl acetate content, grade and additives of the EVA, and we have not been able to confirm it against a materials maker's technical data or standard test methodsNot yet confirmed.

(4) EU Part IV is being revised, and ICH Q3E is a draft

Part IV is a document adopted on 22 November 2017, and in 2025 the EMA issued a concept paper for a revision aligned with Annex 1 [Sources 2 and 16]. We could not confirm the revised version as of this survey (October 2026)Not yet confirmed. ICH Q3E is a draft published in August 2025, with comments invited until 18 December 2025 [Source 13]Not yet confirmed.

(5) Some parts rest on an old review

The mechanisms of freezing, the temperature gradient in liquid nitrogen tanks and the state of cryovials at the time, in chapters 6 and 7, rest on a review from 2011 [Sources 8 and 9]. Tank design and the state of containers may have changed since, and this article has not checked current commercial products.

(6) Primary sources giving numbers on the effect of automation and closed systems are limited

The comparison by Matsumoto et al. is for one process [Source 12]. Within the scope of this survey we could not confirm numbers that generally compare contamination rates between closed and open systems, or show changes in failure rate or cost from automation.

(7) Cost figures are limited rough estimates

The costs of preparing and testing a bank (US$80,000 to US$200,000 and so on) are rough estimates for research grade by a presenter at the 2017 ISCBI workshop [Source 7]. We could not confirm primary sources giving costs for clinical grade or commercial use.

12. Glossary

13. References

1. ICH "Q5D Derivation and Characterisation of Cell Substrates Used for Production of Biotechnological/Biological Products", Step 4, 16 July 1997 https://database.ich.org/sites/default/files/Q5D%20Guideline.pdf
2. European Commission "Guidelines on Good Manufacturing Practice specific to Advanced Therapy Medicinal Products" (EudraLex Volume 4 Part IV), 22 November 2017 https://health.ec.europa.eu/system/files/2017-11/2017_11_22_guidelines_gmp_for_atmps_0.pdf
3. European Commission "EudraLex Volume 4, Annex 1: Manufacture of Sterile Medicinal Products", 25 August 2022 https://health.ec.europa.eu/system/files/2022-08/20220825_gmp-an1_en_0.pdf
4. US FDA (CBER) "Chemistry, Manufacturing, and Control (CMC) Information for Human Gene Therapy Investigational New Drug Applications (INDs)", January 2020 https://www.fda.gov/media/113760/download
5. Director-General of the Pharmaceutical Safety Bureau, MHLW Notification Iyaku-hatsu 0117 No. 11 (17 January 2025), partial revision of the notification on ensuring the quality and safety of human (allogeneic) iPS(-like) cell-processed pharmaceuticals (annexed guideline; posted on the PMDA site) (in Japanese) https://www.pmda.go.jp/files/000273143.pdf
6. CiRA Foundation (Kyoto University iPS Cell Research Foundation) "iPS cell stock project: ten-year summary paper published" (in Japanese), 17 November 2022 https://www.cira-foundation.or.jp/j/news/2022/11/17-000302.html
7. Kim J-H et al. (ISCBI) "A Report from a Workshop of the International Stem Cell Banking Initiative, Held in Collaboration of Global Alliance for iPSC Therapies and the Harvard Stem Cell Institute, Boston, 2017", Stem Cells 37:1130-1135 (2019) https://pmc.ncbi.nlm.nih.gov/articles/PMC7187460/
8. Hunt CJ "Cryopreservation of Human Stem Cells for Clinical Application: A Review", Transfusion Medicine and Hemotherapy 38:107-123 (2011) https://pmc.ncbi.nlm.nih.gov/articles/PMC3088734/
9. Woods EJ, Thirumala S "Packaging Considerations for Biopreservation", Transfusion Medicine and Hemotherapy 38:149-156 (2011) https://pmc.ncbi.nlm.nih.gov/articles/PMC3088737/
10. Tedder RS et al. "Hepatitis B transmission from contaminated cryopreservation tank", The Lancet 346:137-140 (1995) (PubMed abstract) https://pubmed.ncbi.nlm.nih.gov/7603227/
11. Dobruskin M, Toner G, Kander R "Cryopreservation practices in clinical and preclinical iPSC-based cell therapies: Current challenges and future directions", Biotechnology Progress 41 (2025) https://pmc.ncbi.nlm.nih.gov/articles/PMC12348306/
12. Matsumoto E et al. "Fabricating retinal pigment epithelial cell sheets derived from human induced pluripotent stem cells in an automated closed culture system for regenerative medicine", PLoS ONE 14:e0212369 (2019) https://pmc.ncbi.nlm.nih.gov/articles/PMC6415881/
13. ICH (posted by the EMA) "ICH Q3E Guideline for extractables and leachables", Step 2b draft, published 18 August 2025 (EMA/CHMP/ICH/236669/2025) https://www.ema.europa.eu/en/documents/scientific-guideline/ich-q3e-guideline-extractables-leachables_en.pdf
14. PubChem (US National Library of Medicine) "Dimethyl Sulfoxide" (CID 679) https://pubchem.ncbi.nlm.nih.gov/compound/679
15. Lovelock JE, Bishop MW "Prevention of Freezing Damage to Living Cells by Dimethyl Sulphoxide", Nature 183:1394-1395 (1959) https://doi.org/10.1038/1831394a0
16. EMA (GMP/GDP Inspectors Working Group) "Concept paper on the revision of Part IV Guidelines on Good Manufacturing Practice specific to Advanced Therapy Medicinal Products", 11 March 2025 (EMA/INS/GMP/48771/2025) https://www.ema.europa.eu/en/documents/scientific-guideline/concept-paper-revision-part-iv-guidelines-good-manufacturing-practice-specific-advanced-therapy-medicinal-products_en.pdf

14. Claim-to-source audit

Last updated 1 October 2026. Sources are limited to primary material (ICH guidelines, EU Commission GMP guidelines, FDA guidance, MHLW notifications, publications from the CiRA Foundation, and peer-reviewed papers). Because the article includes readings from the materials and process side, those are marked as Commentary and kept separate from sourced fact. Parts on the mechanisms of freezing and the state of containers rest on a 2011 review and may differ from current commercial products and tank designs. A revised EU Part IV and a final ICH Q3E could not be confirmed as of this survey. Market-size and share estimates are not used. This article explains the mechanisms of manufacture and quality control and does not discuss the efficacy or safety of individual products or the choice of treatment. Figs. 1 to 5 are vector drawings; the hero image and Fig. 6 are AI-generated; none of them shows a real facility, product or measured value.